starfish: scalable pipelines for image-based transcriptomics

starfish is a Python library for processing images generated by microscopy-based spatial transcriptomics assays. It lets biologists build scalable pipelines that localize and quantify RNA transcripts in image data generated by any hybridization- or sequencing-based in situ transcriptomics method, from classic RNA single-molecule FISH to combinatorial barcoded assays. Image processing of an experiment is divided into fields of view (FOV) that correspond to the data produced by a microscope at a single location on a microscope slide. starfish lets users register and pre-process images in each FOV, localize spots representing tagged RNA molecules in 3D, decode the identity of those molecules according to the experimental design, segment cells, assign the spots to cells, then aggregate spots into a cell x gene expression matrix. This spatially-annotated gene expression matrix can then be analyzed and visualized in downstream tools for single-cell biology, such as Seurat, Bioconductor, Scanpy, and cellxgene.